Paths Following Algorithm for Penalized Logistic Regression Using SCAD and MCP

In this article, we develop a generalized penalized linear unbiased selection (GPLUS) algorithm. The GPLUS is designed to compute the paths of penalized logistic regression based on the smoothly clipped absolute deviation (SCAD) and the minimax concave penalties (MCP). The main idea of the GPLUS is to compute possibly multiple local minimizers at individual penalty levels by continuously tracing the minimizers at different penalty levels. We demonstrate the feasibility of the proposed algorithm in logistic and linear regression. The simulation results favor the SCAD and MCP’s selection accuracy encompassing a suitable range of penalty levels.     

The Sparse Laplacian Shrinkage Estimator for High-Dimensional Regression

We propose a new penalized method for variable selection and estimation that explicitly incorporates the correlation patterns among predictors. This method is based on a combination of the minimax concave penalty and Laplacian quadratic associated with a graph as the penalty function. We call it the sparse Laplacian shrinkage (SLS) method. The SLS uses the minimax concave penalty for encouraging sparsity and Laplacian quadratic penalty for promoting smoothness among coefficients associated with the correlated predictors. The SLS has a generalized grouping property with respect to the graph represented by the Laplacian quadratic. We show that the SLS possesses an oracle property in the sense that it is selection consistent and equal to the oracle Laplacian shrinkage estimator with high probability. This result holds in sparse, high-dimensional settings with p ? n under reasonable conditions. We derive a coordinate descent algorithm for computing the SLS estimates. Simulation studies are conducted to evaluate the performance of the SLS method and a real data example is used to illustrate its application.     

Sparse optimization for nonconvex group penalized estimation

We consider a linear regression model where there are group structures in covariates. The group LASSO has been proposed for group variable selections. Many nonconvex penalties such as smoothly clipped absolute deviation and minimax concave penalty were extended to group variable selection problems. The group coordinate descent (GCD) algorithm is used popularly for fitting these models. However, the GCD algorithms are hard to be applied to nonconvex group penalties due to computational complexity unless the design matrix is orthogonal. In this paper, we propose an efficient optimization algorithm for nonconvex group penalties by combining the concave convex procedure and the group LASSO algorithm. We also extend the proposed algorithm for generalized linear models. We evaluate numerical efficiency of the proposed algorithm compared to existing GCD algorithms through simulated data and real data sets.     

COORDINATE DESCENT ALGORITHMS FOR NONCONVEX PENALIZED REGRESSION, WITH APPLICATIONS TO BIOLOGICAL FEATURE SELECTION

A number of variable selection methods have been proposed involving nonconvex penalty functions. These methods, which include the smoothly clipped absolute deviation (SCAD) penalty and the minimax concave penalty (MCP), have been demonstrated to have attractive theoretical properties, but model fitting is not a straightforward task, and the resulting solutions may be unstable. Here, we demonstrate the potential of coordinate descent algorithms for fitting these models, establishing theoretical convergence properties and demonstrating that they are significantly faster than competing approaches. In addition, we demonstrate the utility of convexity diagnostics to determine regions of the parameter space in which the objective function is locally convex, even though the penalty is not. Our simulation study and data examples indicate that nonconvex penalties like MCP and SCAD are worthwhile alternatives to the lasso in many applications. In particular, our numerical results suggest that MCP is the preferred approach among the three methods.     

A coordinate descent algorithm for computing penalized smooth quantile regression

The computation of penalized quantile regression estimates is often computationally intensive in high dimensions. In this paper we propose a coordinate descent algorithm for computing the penalized smooth quantile regression (cdaSQR) with convex and nonconvex penalties. The cdaSQR approach is based on the approximation of the objective check function, which is not differentiable at zero, by a modified check function which is differentiable at zero. Then, using the maximization-minimization trick of the gcdnet algorithm (Yang and Zou in, J Comput Graph Stat 22(2):396–415, 2013), we update each coefficient simply and efficiently. In our implementation, we consider the convex penalties \(\ell _1+\ell _2\) and the nonconvex penalties SCAD (or MCP) \(+ \ell _2\). We establishe the convergence property of the csdSQR with \(\ell _1+\ell _2\) penalty. The numerical results show that our implementation is an order of magnitude faster than its competitors. Using simulations we compare the speed of our algorithm to its competitors. Finally, the performance of our algorithm is illustrated on three real data sets from diabetes, leukemia and Bardet–Bidel syndrome gene expression studies.     

High-dimensional grouped folded concave penalized estimation via the LLA algorithm

The group folded concave penalization problems have been shown to process the satisfactory oracle property theoretically. However, it remains unknown whether the optimization algorithm for solving the resulting nonconvex problem can find such oracle solution among multiple local solutions. In this paper, we extend the well-known local linear approximation (LLA) algorithm to solve the group folded concave penalization problem for the linear models. We prove that, with the group LASSO estimator as the initial value, the two-step LLA solution converges to the oracle estimator with overwhelming probability, and thus closing the theoretical gap. The results are high-dimensional which allow the group number to grow exponentially, the true relevant groups and the true maximum group size to grow polynomially. Numerical studies are also conducted to show the merits of the LLA procedure.     

APPLE: approximate path for penalized likelihood estimators

In high-dimensional data analysis, penalized likelihood estimators are shown to provide superior results in both variable selection and parameter estimation. A new algorithm, APPLE, is proposed for calculating the Approximate Path for Penalized Likelihood Estimators. Both convex penalties (such as LASSO) and folded concave penalties (such as MCP) are considered. APPLE efficiently computes the solution path for the penalized likelihood estimator using a hybrid of the modified predictor-corrector method and the coordinate-descent algorithm. APPLE is compared with several well-known packages via simulation and analysis of two gene expression data sets.     

NETWORK EXPLORATION VIA THE ADAPTIVE LASSO AND SCAD PENALTIES

Graphical models are frequently used to explore networks, such as genetic networks, among a set of variables. This is usually carried out via exploring the sparsity of the precision matrix of the variables under consideration. Penalized likelihood methods are often used in such explorations. Yet, positive-definiteness constraints of precision matrices make the optimization problem challenging. We introduce non-concave penalties and the adaptive LASSO penalty to attenuate the bias problem in the network estimation. Through the local linear approximation to the non-concave penalty functions, the problem of precision matrix estimation is recast as a sequence of penalized likelihood problems with a weighted L(1) penalty and solved using the efficient algorithm of Friedman et al. (2008). Our estimation schemes are applied to two real datasets. Simulation experiments and asymptotic theory are used to justify our proposed methods.     

The cluster correlation-network support vector machine for high-dimensional binary classification

ABSTRACT

Identifying homogeneous subsets of predictors in classification can be challenging in the presence of high-dimensional data with highly correlated variables. We propose a new method called cluster correlation-network support vector machine (CCNSVM) that simultaneously estimates clusters of predictors that are relevant for classification and coefficients of penalized SVM. The new CCN penalty is a function of the well-known Topological Overlap Matrix whose entries measure the strength of connectivity between predictors. CCNSVM implements an efficient algorithm that alternates between searching for predictors’ clusters and optimizing a penalized SVM loss function using Majorization–Minimization tricks and a coordinate descent algorithm. This combining of clustering and sparsity into a single procedure provides additional insights into the power of exploring dimension reduction structure in high-dimensional binary classification. Simulation studies are considered to compare the performance of our procedure to its competitors. A practical application of CCNSVM on DNA methylation data illustrates its good behaviour.     

Bridge regression: Adaptivity and group selection

In high-dimensional regression problems regularization methods have been a popular choice to address variable selection and multicollinearity. In this paper we study bridge regression that adaptively selects the penalty order from data and produces flexible solutions in various settings. We implement bridge regression based on the local linear and quadratic approximations to circumvent the nonconvex optimization problem. Our numerical study shows that the proposed bridge estimators are a robust choice in various circumstances compared to other penalized regression methods such as the ridge, lasso, and elastic net. In addition, we propose group bridge estimators that select grouped variables and study their asymptotic properties when the number of covariates increases along with the sample size. These estimators are also applied to varying-coefficient models. Numerical examples show superior performances of the proposed group bridge estimators in comparisons with other existing methods.     

An Algorithm for Unconstrained Quadratically Penalized Convex Optimization

Estimators are often defined as the solutions to data dependent optimization problems. A common form of objective function (function to be optimized) that arises in statistical estimation is the sum of a convex function V and a quadratic complexity penalty. A standard paradigm for creating kernel-based estimators leads to such an optimization problem. This article describes an optimization algorithm designed for unconstrained optimization problems in which the objective function is the sum of a non negative convex function and a known quadratic penalty. The algorithm is described and compared with BFGS on some penalized logistic regression and penalized L ^3/2 regression problems.     

An elastic-net penalized expectile regression with applications

To perform variable selection in expectile regression, we introduce the elastic-net penalty into expectile regression and propose an elastic-net penalized expectile regression (ER-EN) model. We then adopt the semismooth Newton coordinate descent (SNCD) algorithm to solve the proposed ER-EN model in high-dimensional settings. The advantages of ER-EN model are illustrated via extensive Monte Carlo simulations. The numerical results show that the ER-EN model outperforms the elastic-net penalized least squares regression (LSR-EN), the elastic-net penalized Huber regression (HR-EN), the elastic-net penalized quantile regression (QR-EN) and conventional expectile regression (ER) in terms of variable selection and predictive ability, especially for asymmetric distributions. We also apply the ER-EN model to two real-world applications: relative location of CT slices on the axial axis and metabolism of tacrolimus (Tac) drug. Empirical results also demonstrate the superiority of the ER-EN model.     

Variable selection for longitudinal data with high-dimensional covariates and dropouts

A new variable selection approach utilizing penalized estimating equations is developed for high-dimensional longitudinal data with dropouts under a missing at random (MAR) mechanism. The proposed method is based on the best linear approximation of efficient scores from the full dataset and does not need to specify a separate model for the missing or imputation process. The coordinate descent algorithm is adopted to implement the proposed method and is computational feasible and stable. The oracle property is established and extensive simulation studies show that the performance of the proposed variable selection method is much better than that of penalized estimating equations dealing with complete data which do not account for the MAR mechanism. In the end, the proposed method is applied to a Lifestyle Education for Activity and Nutrition study and the interaction effect between intervention and time is identified, which is consistent with previous findings.     

Lasso penalized semiparametric regression on high-dimensional recurrent event data via coordinate descent

This paper studies a fast computational algorithm for variable selection on high-dimensional recurrent event data. Based on the lasso penalized partial likelihood function for the response process of recurrent event data, a coordinate descent algorithm is used to accelerate the estimation of regression coefficients. This algorithm is capable of selecting important predictors for underdetermined problems where the number of predictors far exceeds the number of cases. The selection strength is controlled by a tuning constant that is determined by a generalized cross-validation method. Our numerical experiments on simulated and real data demonstrate the good performance of penalized regression in model building for recurrent event data in high-dimensional settings.     

A coordinate majorization descent algorithm for ℓ1 penalized learning

The glmnet package by Friedman et al. [Regularization paths for generalized linear models via coordinate descent, J. Statist. Softw. 33 (2010), pp. 1–22] is an extremely fast implementation of the standard coordinate descent algorithm for solving ?₁ penalized learning problems. In this paper, we consider a family of coordinate majorization descent algorithms for solving the ?₁ penalized learning problems by replacing each coordinate descent step with a coordinate-wise majorization descent operation. Numerical experiments show that this simple modification can lead to substantial improvement in speed when the predictors have moderate or high correlations.     

Marginalized lasso in sparse regression

We propose marginalized lasso, a new nonconvex penalization for variable selection in regression problem. The marginalized lasso penalty is motivated from integrating out the penalty parameter in the original lasso penalty with a gamma prior distribution. This study provides a thresholding rule and a lasso-based iterative algorithm for parameter estimation in the marginalized lasso. We also provide a coordinate descent algorithm to efficiently optimize the marginalized lasso penalized regression. Numerical comparison studies are provided to demonstrate its competitiveness over the existing sparsity-inducing penalizations and suggest some guideline for tuning parameter selection.     

Quadratic Programming and Penalized Regression

Quadratic programming is a versatile tool for calculating estimates in penalized regression. It can be used to produce estimates based on L ₁ roughness penalties, as in total variation denoising. In particular, it can calculate estimates when the roughness penalty is the total variation of a derivative of the estimate. Combining two roughness penalties, the total variation and total variation of the third derivative, results in an estimate with continuous second derivative but controls the number of spurious local extreme values. A multiresolution criterion may be included in a quadratic program to achieve local smoothing without having to specify smoothing parameters.     

Removing the singularity of a penalty via thresholding function matching

By introducing the idea of thresholding function matching, it is illustrated that both bridge penalty and log penalty can be transformed so as to circumvent certain difficulties in numerical computation and the definition of local minimality. The fact that both bridge penalty and log penalty have derivatives going to infinity at zero. This hinders their applications in statistics although it is reported in the literature that they allow recovery of sparse structure in the data under some conditions. It is illustrated in the simulation studies that in the variable selection problems, penalized likelihood estimation based on the transformed penalty obtained by the proposed thresholding function matching method outperform those based on many other state-of-art penalties, particularly when the covariates are strongly correlated. The one-to-one correspondence between the transformed penalties and their thresholding functions are also established.     

Approximate penalization path for smoothly clipped absolute deviation

Feature selection often constitutes one of the central aspects of many scientific investigations. Among the methodologies for feature selection in penalized regression, the smoothly clipped and absolute deviation seems to be very useful because it satisfies the oracle property. However, its estimation algorithms such as the local quadratic approximation and the concave–convex procedure are not computationally efficient. In this paper, we propose an efficient penalization path algorithm. Through numerical examples on real and simulated data, we illustrate that our path algorithm can be useful for feature selection in regression problems.     

Quadratic Approximation via the SCAD Penalty with a Diverging Number of Parameters

The high-dimensional data arises in diverse fields of sciences, engineering and humanities. Variable selection plays an important role in dealing with high dimensional statistical modelling. In this article, we study the variable selection of quadratic approximation via the smoothly clipped absolute deviation (SCAD) penalty with a diverging number of parameters. We provide a unified method to select variables and estimate parameters for various of high dimensional models. Under appropriate conditions and with a proper regularization parameter, we show that the estimator has consistency and sparsity, and the estimators of nonzero coefficients enjoy the asymptotic normality as they would have if the zero coefficients were known in advance. In addition, under some mild conditions, we can obtain the global solution of the penalized objective function with the SCAD penalty. Numerical studies and a real data analysis are carried out to confirm the performance of the proposed method.